FINAL REPORT For: TerraSystems, Inc. Jonathan Gradie 2800 Woodlawn Drive, Suite 264 Honolulu, HI 96822 Phone: 539-3745 ASSESSING THE AMOUNT OF CORAL GROWTH IN HIGH RECREATIONAL USE AREAS OF KANEOHE BAY A CASE STUDY USING REMOTE SENSING ON REEFS NEAR COCONUT ISLAND By: Christian Werjefelt 277 Kaha St.
24
Embed
FINAL REPORT 539-3745 - University of Hawaii · 2012-08-16 · FINAL REPORT For: TerraSystems, Inc. Jonathan Gradie 2800 Woodlawn Drive, Suite 264 Honolulu, HI 96822 Phone: 539-3745
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
FINAL REPORT
For: TerraSystems, Inc. Jonathan Gradie
2800 Woodlawn Drive, Suite 264 Honolulu, HI 96822
Phone: 539-3745
ASSESSING THE AMOUNT OF CORAL GROWTH IN HIGH RECREATIONAL USE AREAS OF KANEOHE BAY
A CASE STUDY USING REMOTE SENSING ON REEFS NEAR COCONUT ISLAND
By: Christian Werjefelt
277 Kaha St. Kailua, Hawaii 96734
Phone: 261-8893
December 18, 1996
University of Hawaii MOPS Certificate J
Sherwood Maynard, MOPS advisor Phone: 956-8433
Fax: 956-2417
MARINE OPTION PROGRAM
ABSTRACT
The objective of this project was to assess the coral health in areas
around Coconut Island by comparing a high use recreation area versus a low
use area a s a means of measuring the impacts of human activity on the reef sys-
tem. My plan to accomplish this project was to use multi-spectral imagery
gathered from a n airplane of the area surrounding Coconut island. This da ta
was collected by Terrasystems Inc. (TSI) as part of a much larger survey of the
coral health in all of Kaneohe Bay headed by Marlin Atkinson and Mr. Eric
Hochberg of the School of Ocean and Earth Science Technology (SOEST) at the
University of Hawaii. This comprehensive survey of Kaneohe Bay by SOEST is
part of an ongoing project to determine the methodology of using remote sens-
ing, in particular spectral imagery, to map coral reefs worldwide from satellites.
Kaneohe Bay is being used as a test site of this technology. My project repre-
sents a n offshoot of this comprehensive project by SOEST. The results of this
project indicate little coral growth near the eco-tourist site. However, a t this
time there is no distinct correlation between the lack of coral and the tourists.
Further studies need to be done in order to find out if the tourists a re impacting
the coral reef. In addition, the results show tha t spectral imagery will be a n
important tool for mapping and understanding coral reef health in high recre-
ational use areas.
INTRODUCTION
KANEOHE BAY
Kaneohe Bay is located on the windward side of the island of Oahu (Fig.
1). The bay is protected by a barrier reef which has created a sheltered envi-
ronment where corals grow and flourish (Laws, 1993). The corals play a n
Figure 1 Oahu
Kaneohe Bay and environs.
- Kaneohe Bay
Sources: Oahu Map: Morgan 1996 Kaneohe Bay Map: Laws 1993
important par t in the survival of the ecosystem tha t has evolved in the bay.
They supply food and shelter for the organisms, such as fish and invertebrates,
and help to maintain the ecology within the bay.
I n the past, Kaneohe Bay had sewage pumped into the southeast sector of
the bay. This led to the degradation of much of the coral in the affected area.
The discharge lasted for about thirty years, ending in 1978. The discharge was
ended because of the effects to the corals and since then the coral has been
growing back (Laws, 1993).
Researchers a t SOEST want to study how low earth orbit satellites can be
used for monitoring coral reefs world wide. This proposed system, which is
being tested in concept using Kaneohe Bay, could be used to map coral destruc-
tion and health, as well as monitoring the recovery of the coral. This technology
could also be used to identify areas for recreational use, or to assess the s ta tus
of boat channels, in addition to serving as a worldwide model for future projects
regarding multi-spectral imagery of corals. Eric Hochberg (graduate student,
SOEST) will be using the entire multi-spectral data set to create a n entire map
of the reefs in Kaneohe Bay. This map will show areas of coral, sand, algae, and
deep water. The goal is to show that this technology can be transferred to satel-
lite remote sensing (Hochberg and Atkinson, in prep.).
For my project I wanted to study the difference in the amount of coral
growth between an area with high human use and an area with little use. I
decided to use the area around and near Coconut Island (Fig. 2), in the south-
east sector of the bay, for my study area. This was an ideal location because the
coral surrounding Coconut Island is protected and has very little exposure to
human contact and destruction. However, one of the patch reefs near the
island, known a s Checker Reef, gets heavy use because i t is used as a n eco-
Figure 2
Photo of Coconut Island in Kaneohe Bay
tourism site. Tourists are brought to the reef to enjoy snorkeling, jet skiing,
and wind surfing. This project proposes to assess how much damage is being
done by the tourists to the coral in the Checker Reef area.
REMOTE SENSING
Remote sensing is very useful for obtaining knowledge of the characteris-
tics in a certain area. In general, remote sensing uses electromagnetic radia-
tion sensors to record images of the environment which can be interpreted to
yield useful information of an area. Remote sensing provides information by
measuring the intensity of the electromagnetic radiation being reflected or emit-
ted from the earth's surface. Measures of the electromagnetic radiation, as a
function of wavelength, give information concerning the shape, topography, and
composition of the ground (Curran, 1985). The information gained can be put to
use in all areas concerning environmental sciences.
One area where the application of remote sensing is beneficial is in the
management of our natural resources. For example, it has been used in forestry
to monitor the effects of logging. Also, geologists can use remote sensing to sur-
vey the surface features of an area to find places conducive to mining (Warner,
e t al., 1996). Furthermore, remote sensing can also be beneficial in monitoring
agricultural land and vegetation cover (Drury, 1990).
Remote sensing is commonly accomplished through the use of satellites or
airplanes. However, other means can be used such a s a boat. For this project I
will be analyzing data collected from a n airplane using multi-spectral imagery.
Multi-spectral imagery uses specific wavelengths to take digital images of the
earth. The information is obtained by reading how strong the reflectance of
each band is. This gives the information needed to analyze the spatial a rea
examined and its composition (Drury, 1990).
Data collected from planes must be gathered very systematically. To do
this, the area examined is divided into evenly spaced parallel lines, known as
'flight paths' (Fig. 3) . The plane follows each flight path while the multi spec-
t ra l scanner takes pictures periodically to cover the entire area (Gradie, pers.
comm.).
PROCEDURES
INFORMATION GATHERING ON REMOTE SENSING
I reviewed literature a t Hamilton Library on remote sensing and i t s
applications. I found several useful books: fundl F o r r u L L h i d Photoprzuhy
(Warner, e t al., 1996), P r inc i~ les of Remote Sensing (Curran, 1985), and A Guide
$0 Remote Sensing (Drury, 1985). This has helped to familiarize me with remote
sensing and i ts uses.
I n addition, I have also used the internet as a n information source. The
SOEST program has a web site entitled 'Virtually Hawaii' which can be located
at: http:\\www.soest.hawaii.edu/. This contains many remotely sensed images
of Hawaii along with a tutorial about remote sensing which has also been very
helpful in gaining a foundation of knowledge about remote sensing.
I n addition, I have found some information concerning previous work
using remote sensing for coral reefs. One study in the Dominican Republic used
Landsat da ta from a satellite to distinguish between coral, seagrass, and sandy
areas. This project was partly successful in tha t some areas were able to be
mapped while others were not. Specifically, i t was easy to distinguish sand
from coral and seagrass but not the seagrass from coral. The reason for this is
because the seagrass and coral reflect similar amounts of specific wavelengths
Figure 3
Flight Paths over Kaneohe Bay: October 1996
Note: Dark horizontal lines represent flight paths taken by TSI.
Source: TSI
whereas sand does not. Sand has a higher reflectance and had a significantly
greater mean Landsat digital count than did seagrass or coral, making i t easier
to distinguish. One problem with this project was tha t because the da ta was
from a satellite, the pixel size used was very large a t 812 m2 (Luczkovich e t a1
1993). This could present problems because within such a large area there a re
many small patches made up of varying bottom types. A pixel in such a n area
would not give a clear indication of a specific bottom type.
Another study was done on Heron Reef, located within the Great Barrier Reef
Marine Park, to see how well remote sensing could be used to determine differences in
coral reef zones. Fourteen reef zone groups were used. These zones varied according to
depth andlor bottom type and ranged from open water to breaking waves and from corals
to sand (Ahmad and Neil 1994). This study was quite successful and shows that coral
density gradients and distribution patterns can be found using remote sensing.
Damage to coral reef areas from direct human contact often occurs. A
study done of the reef-flats a t Sharm-el-Sheikh (a popular diving and tourist
a rea on the Egyptian Red Sea), shows tha t damage to the coral has occurred
from divers and tourists. The results indicate that there is significantly more
coral damage, more rubble and bare rock, and smaller coral colonies in heavily
trampled areas than in less trampled areas (Hawkins 1993).
Although damages to coral reefs from direct human contact are evident
there did not seem to be any data showing an attempt to use remote sensing to
specifically measure tha t damage.
I also contacted and met with Eric Hochberg who is currently doing work
in Kaneohe Bay using remote sensing. This project is being done by the SOEST
department a t UH with Dean Barry Raleigh and Marlin Atkinson in charge.
Mr. Hochberg introduced me to his project which involves creating a n
entire map of Kaneohe Bay and its reefs through the use of remote sensing. He
helped to explain the concept of using band ratios to discern different materials
on the reef. Mr. Hochberg's work has formed the basis of my project because he
has been able to identify which multi-spectral bands are most useful in the map-
ping of a reef. Four bands were selected by Mr. Hochberg based upon hyper-
spectral imagery, acquired by SETS Technology Inc., of a small portion of the
reef system which Eric previously studied. This information was used by TSI to
gather the relevant data of Kaneohe Bay.
MULTI SPECTRAL DATA GATHERING OF KANEOHE BAY
Data has been gathered by TSI using aerial multi-spectral imagery. Four
narrow bands a t 488nm, 550nm, 570nm, and 701nm were used to do this. As
mentioned earlier, these bands were chosen by Mr. Hochberg after careful study
of previous hyperspectral data gathered of the area.
TSI took 263 images of Kaneohe Bay on October 5, 1996. The plane flew
a t 10,000 feet and used fourteen different flight paths to cover the bay. As the
plane flew along, an image was taken every 7 seconds. Each image is made up
of tiny squares known as pixels which, for this data, are two square meters.
Each image is 578 pixels by 740 pixels (Figure 4).
CORRECTING DATA
During the data collection process there are several artifacts which arise
tha t need to be adjusted before data can be analyzed. These artifacts come
about from a variety of reasons including inadvertent electrical interference and
image misalignment. Image misalignment arises due to the movement of the
plane or because the lenses of the camera are not in the same place when the
Figure 4
pixels image
r d L
740 pixels (1480 m) .
one pixel is 2 square meters
image is taken (Gradie, pers. comm.).
Electrical interference can induce a regular pattern of spots on a n image.
This interference is known as 'noise' and should be removed for the best inter-
pretation of the image. Removing the 'noise' is known as 'noise reduction' using
a process called Fourier Transforms (Gradie, pers. comm.). This process is com-
plex and difficult to understand. Although 'noise' is not always a problem, I was
introduced to this concept in order to understand i t s effects.
An artifact normally encountered is a condition known as 'offset'. This
arises due to the movement of the plane which creates spatial differences
between the separate bands. Luckily, this can be easily corrected on the com-
puter through a simple but tedious process known as 'offsetting' (Gradie, pers.
comm.).
Another artifact which must be addressed on all images arises from the
spatial misalignment of the camera system and its lenses. This occurs because
the four lenses used are not aimed a t exactly the same point on the ground
when the image is taken. To correct this problem the images must be 'co-regis-
tered' together. This process uses one band as the base image and rotates,
stretches, and translates the other three bands to match the spatial orientation
of the base image (Gradie, pers. comm.). This creates a nice clear image. This
work was done on the ENVI (Environment for Visualizing Images) program.
ENVI is the program used a t TSI to read, analyze and process remotely sensed
data. I t also allows maps to be made from the images.
SELECTION OF SITE
There were two different areas tha t I chose as possibilities to study for
this project. One area was the southeast end of Kaneohe Bay where the
Kaneohe Yacht Club is located and the other was the area around Coconut
Island.
The area near Kaneohe Yacht Club is interesting because of the amount of
use and human contact this section of the bay gets. There are many boats going
in and out of the yacht club a t all times and homes line the shoreline which
increases human interaction with the bay. In addition, this end of the bay is
where the old sewage outfall was. I feel i t would be interesting to assess how
much coral there currently is in the area. However, the images taken in this
area had some cloud cover which could make the process of analyzing the data
more difficult. For this reason I did not choose this area to study.
Coconut Island is also an interesting area with high usage. Many boats
travel around and to the island. In addition, a reefs near the island named
Checker Reef is used for eco-tourism purposes. Boats from Heeia Kea boat har-
bor bring tourists to the reef to enjoy the waters of Kaneohe Bay. They are
allowed to snorkel, wind surf, and jet ski. I thought i t would be interesting to
see if there are any effects to the reefs in the area due to the interaction of the
tourists. The images collected near Coconut Island were very clear and there-
fore I chose this area to study rather than Kaneohe Yacht Club. The project area
is shown on the following page (Fig. 5) with Coconut Island, Checker Reef, and
the recreation area.
MOSAICING THE IMAGES
In order to get a large area to study, several images must be 'mosaiced'
together. This is done by overlapping the edges of two images to align them
according to landmarks on both the images (Figure 6). This is possible because
when the images are taken from the airplane the flight paths overlap one anoth-
I P
Figure 5
Figure 6
Note: -Representation of mosaicing two images together -Oval shape represents an island
er slightly to insure that all points on the ground are covered (Gradie, pers.
comm.).
Five images have been mosaiced together to create the clear picture of the
entire Coconut Island area and its surroundings (Fig. 5).
GEOREFERENCING THE IMAGE
'Georeferencing' is the process whereby the area to be analyzed serves as
an overlay to a digitized map. This is done by selecting the same points on the
image as on the map to match with one another. This creates latitudinal and
longitudinal degrees on the observed area. Thus, the image is 'georeferenced'.
This allows for accurate identification of locations in the area. In order to do
this, a map has to be selected and digitally scanned before i t can be used
(Gradie, pers. comm. 1.
ANALYZING THE DATA
To analyze the data for this project, band ratios were used. Band ratios
allow differences in composition of materials to be seen more readily. They are
created by dividing one band with another.
RESULTS
TSI was successful in collecting data needed for the SOEST project. This
data covered the entire area of Kaneohe Bay using 263 images. My project
entails the study of only one of these images, centered near Coconut Island.
Prior to the analysis, corrections were made to the data as needed. These cor-
rections included co-registration, offsetting, and georeferencing (discussed in the
procedures section).
A small study section in the area where the tourists are on Checker Reef
was chosen to be analyzed along with an area on the reef around Coconut Island
(Fig. 7). I t was simpler to do the analysis in this manner. Reasons for this are
explained in the 'discussion' portion of the paper.
The analyzed sample area of Coconut Island (Fig. 8) consists of four col-
ors. Each color represents a difference in spectral characteristics or a difference
in material. The meaning of each color is as follows: yellow represents sand,
green represents algae, red represents living coral, and blue represents deep
water (water below approximately 5 m). The area which extends approximately
to the 150 m mark in each direction is Coconut Island. The analysis indicates
that the island is a combination of these colors. This area should be disregard-
ed. The pertinent area is the perimeter of the reef (the area which meets the
deep water). These preliminary results indicate a high amount of coral growth
along the perimeter. There is also some sand and a lot of algae.
The analyzed sample of the Checker Reef recreation area (Fig. 9) consists
of the same colors representing the same materials as the Coconut Island sam-
ple: yellow represents sand, green represents algae, red represents living coral,
and blue represents deep water (below approximately 5 m). The areas in the
deep water, which show up as mostly coral and some sand, are actually the
boats and floaters from the tourist operations. These areas should be disregard-
ed as they are not accurate. However, the pertinent area is the perimeter of the
reef (the area which meets the deep water). According to the results there is lit-
tle coral growth directly in front of the tourist area. This area consists mainly
of sand and algae. However, there is some coral growth off to the sides.
Figure 7
-Shows the approximate areas sampled from checker reef and coconut island.
Figure 8 Coconut Island Sam~le Area
50 100 150 200 250 300 Distances in meters
Blue- Deep Water Yellow- Sand Green- Algae u- Living Coral
Figure 9 Checker Reef Sample Area
0 50 100 150 200 250 300 Distances in meters
Blue- Deep Water Yellow- Sand Green- Algae u- Live Coral
DISCUSSION
Unfortunately, up to this point, data analysis has not been completely
successful. This is because the algorithms, which Eric Hochberg developed from
a previous study, didn't work very well with the new data gathered. However,
Mr. Hochberg and Professor Atkinson will most likely be able to complete their
project of mapping Kaneohe Bay once they have been able to make corrections
needed for the algorithms.
The previous study done by Mr. Hochberg used hyperspectral remote sens-
ing to survey a small area of Kaneohe Bay prior to the project with TSI.
Hyperspectral remote sensing is not limited to a few bands but encompasses all
wavelengths. With this data Mr. Hochberg was able to choose four bands
(488nm, 550nm, 570nm, 701nm) to simplify his project with TSI. He then devel-
oped a n algorithm for the area using the four bands to discern between different
materials on the reef. He was able to distinguish areas of coral, algae, and sand
or rubble. However, this algorithm has not been as successful in discerning
materials using the new data gathered by TSI for all of Kaneohe Bay.
Due to the problems with the algorithms, small study areas were chosen
instead of studying the entire area. Mr. Hochberg was able to analyze these
da ta samples using his algorithms to come up with the two diagrams shown of
the areas. However, i t must be remembered that these data sets a re prelimi-
nary and should be tested again after the algorithms have been perfected to
insure their accuracy.
Although the findings indicate that there is little coral growth in the
study area of Checker Reef, it cannot be assumed this is specifically due to the
effects of the tourists. In order to accurately assess this, one would have to
know the amount of coral which existed on Checker Reef prior to the eco-
tourism. Had there been remotely sensed data of the area prior to the eco-
tourism, this could be assessed. The area on Checker Reef may have always
been sand and was chosen as an eco-tourism location because of this.
As mentioned in the results, both areas sampled contain items which are
above water. The Checker Reef area contains what are believed to be boats and
buoys and the Coconut Island sample has a portion of the island in it. These
areas above water show up as combinations of different materials and should be
disregarded. However, i t can be assumed that the area on the outer portions of
the reefs is generally accurate.
There simply was not enough time to accomplish all that I hoped to do for
this project. Interpreting the data will be quite simple once the correct algo-
rithms are calculated. In addition, more research is needed in studying the
area in order to truly assess the amount of damage done and what the damage
stems from.
In the future this data can be used for other projects. For example, if
new remotely sensed data are collected a t a later date, i t can be compared to the
data collected in this project to look a t differences in coral growth. The effects
of eco-tourists on an area can then be more accurately assessed.
Such data would be useful in eco-tourism management of the reefs. I t
would help in determining a sustainable amount of people for an area over a
specific amount of time. This could keep the coral in the area healthy, assuming
that there is some damage done to the area which could be linked to the activi-
ties of the tourists.
In the end, this project was ultimately about learning what remote sens-
ing is rather than coming up with a specific conclusion about the eco-tourism
effects to the reef. I foresaw that such a project would need more time and
research but felt it would be interesting to attempt, It was beneficial to learn
what remote sensing is, how it is used, and how it can be applied. This is
knowledge which will be useful in the future and applies to a wide variety of
environmental fields.
REFERENCES CITED
Ahmad, W., Neil, D.T., (1994). An evaluation of Landsat Thematic Mapper (TM) digital data for discriminating coral reef zonation: Heron Reef (GBR). In: International Journal of Remote Sensing, 15 (13): 2583-2597.
Bainbridge, S. J. S m n s i n ~ of Coral Reefs: An O v e r v i e ~ ; Remote Sensing in the ASEAN Region: An Australian Perspective. 1988.
Curran, Paul J. v o t e m; Longman Inc., New York. 1985.
Drury, S.A. A Guide to Remote Sensing: I n t e r ~ r e t i n y Imaees of the Earth; Oxford University Press. New York. 1990.
Gradie, Jonathan. personal communication. 1996.
Hawkins, Julie F? and Roberts, Callurn M., (1993) Effects of recreational scuba div- ing on coral reefs: trampling on reef-flat communities. In: Journal of Applied Ecology. Vol. 30 p. 25-30.
Hochberg, Eric and Marlin Atkinson. in prep. 1996. (work phone: 956- 5397)
Laws, Edward A. A auatic Pollution: An Introductorv Text; John Wiley & Sons, Inc. United States. 1993.
Luczkovich, Joseph J., Wagner, Thomas W., Michalek, Jeffrey L., StoMe, Richard W.. (1993). Discrimination of Coral Reefs, Seagrass Meadows, and Sand Bottom Qpes from Space: A Dominican Republic Case Study. In: Photogrammetric Engineering and Remote Sensing, Vol. 59, No. 3, March 1993, pp. 385-389.
. . Morgan, Joseph R. Hawai: A Uniaue Geoma~hy. The Bess Press, Inc. Korea. 1996.
SOEST. http:\\www.soest.hawaii.edu/
Warner, W.S., R.W. Graham, R.E. Read; Small Format Aerial Photo~ra*; Interprint Ltd., Malta. 1996.